An object image detection device is disclosed that is able to rapidly detect an object image from an input image without a great deal of computation. The object image detection device includes an object image classification unit for determining whether the object images are included in an image having a given orientation, an image orientation detection unit for detecting orientation of the input image, an image rotation unit for rotating the object image classification unit according to the detected orientation of the input image, and a detection unit for detecting the object images from the input image by using the rotated object image classification unit.
Legal claims defining the scope of protection, as filed with the USPTO.
1. An object image detection device for detecting one or more object images from an input image, comprising: a reference object image classification unit configured to determine whether an object image is included in an image having a predetermined orientation; an image orientation detection unit configured to detect an orientation of the input image; an image rotation unit configured to read parameters used in the reference object image classification unit and an angle of object images of the reference image classification unit and rotate the parameters in response to the angle of the object images to generate an image classification unit adapted to the detected orientation; and a detection unit configured to detect and extract at least one of position information and size information of the object images from the input image based on the generated image classification unit for display on an image output unit.
2. The object image detection device as claimed in claim 1 , wherein the image orientation detection unit detects the orientation of the input image by using one or more features of the object image.
3. The object image detection device as claimed in claim 1 , wherein the object image classification unit includes a plurality of image classifiers arranged in cascade manner, and the image rotation unit rotates at least a part of the image classifiers of the object image classification unit according to the detected orientation of the input image.
4. The object image detection device as claimed in claim 1 , wherein the image orientation detection unit includes a plurality of direction identifiers each able to distinguish two angles from the others.
5. The object image detection device as claimed in claim 1 , wherein the image orientation detection unit includes an integration section configured to integrate outputs of the direction identifiers.
6. The object image detection device as claimed in claim 1 , wherein the image rotation unit rotates the object image classification unit by adjusting direction-sensitive parameters of the object image classification unit according to the detected orientation of the input image.
7. The object image detection device as claimed in claim 6 , wherein the direction-sensitive parameters include positions of features of the object image.
8. An object image detection method for detecting one or more object images from an input image, comprising: inputting an image into an object image detecting device; an image orientation detection step for detecting an orientation of the input image using an image orientation detector; an image rotation step including reading parameters used in the reference object image classification unit and an angle of object images of the reference image classification unit for rotating parameters used in a reference image classification unit in response to the angle of the object images to generate an image classification unit adapted to the detected orientation; and an object image classification step for detecting and extracting at least one of position information and size information of the object image from the input image based on the rotated parameters used in the object image classification.
9. The object image detection method as claimed in claim 8 , wherein in the image orientation detection step, the orientation of the input image is detected by using one or more features of the object image.
10. The object image detection method as claimed in claim 8 , wherein in the object image classification step, a plurality of image classifiers arranged in cascade manner is used to determine whether the object image is included in the input image, and in the image rotation step, at least a part of the image classifiers are rotated according to the detected orientation of the input image.
11. The object image detection method as claimed in claim 8 , wherein in the image orientation detection step, a plurality of direction identifiers each able to distinguish two angles from the others is used to detect the orientation of the input image.
12. The object image detection method as claimed in claim 11 , wherein in the image orientation detection step, outputs of the direction identifiers are integrated.
13. The object image detection method as claimed in claim 8 , wherein in the image rotation step, among the parameters used in the object image classification step, only direction-sensitive parameters are adjusted according to the detected orientation of the input image.
14. The object image detection method as claimed in claim 13 , wherein the direction-sensitive parameters include positions of features of the object images.
15. A non-transitory program product, which when read and executed by a computer, performs an object image detection method for detecting one or more object images from an input image, said method comprising: an image orientation detection step for detecting orientation of the input image; an image rotation step including reading parameters used in the reference object image classification unit and an angle of object images of the reference image classification unit for rotating parameters used in a reference image classification unit in response to the angle of the object images to generate an image classification unit adapted to the detected orientation; and an object image classification step for detecting and extracting at least one of position information and size information of the object image from the input image based on the rotated parameters used in the object image classification.
16. The non-transitory program product as claimed in claim 15 , wherein in the image orientation detection step, the orientation of the input image is detected by using one or more features of the object image.
17. The non-transitory program product as claimed in claim 15 , wherein in the object image classification step, a plurality of image classifiers arranged in cascade manner is used to determine whether the object image is included in the input image, and in the image rotation step, at least a part of the image classifiers are rotated according to the detected orientation of the input image.
18. The non-transitory program product as claimed in claim 15 , wherein in the image orientation detection step, a plurality of direction identifiers each able to distinguish two angles from the others is used to detect the orientation of the input image.
19. The non-transitory program product as claimed in claim 18 , wherein in the image orientation detection step, outputs of the direction identifiers are integrated.
20. The non-transitory program product as claimed in claim 15 , wherein in the image rotation step, among the parameters used in the object image classification step, only direction-sensitive parameters are adjusted according to the detected orientation of the input image.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
March 20, 2008
February 25, 2014
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.